Research Article

A central master driver of psychosocial stress responses in the rat

See allHide authors and affiliations

Science  06 Mar 2020:
Vol. 367, Issue 6482, pp. 1105-1112
DOI: 10.1126/science.aaz4639

A major psychosocial stress circuit

Psychological stress induces various physiological responses by activating the sympathetic nervous system. The brain circuits involved in these functions are still not completely understood. In a rat model, Kataoka et al. combined anatomical tracing, immediate early gene expression analysis, pharmacology, optogenetics, electrophysiology, and genetic cell ablation to provide evidence for the prominent role of a ventral part of the medial prefrontal cortex in sympathetic responses to social defeat stress. This brain region sends excitatory projections to the dorsomedial hypothalamus as a central coordinator of the psychosocial stress responses. This pathway is crucial for understanding how psychosocial stress influences a variety of body functions.

Science, this issue p. 1105


The mechanism by which psychological stress elicits various physiological responses is unknown. We discovered a central master neural pathway in rats that drives autonomic and behavioral stress responses by connecting the corticolimbic stress circuits to the hypothalamus. Psychosocial stress signals from emotion-related forebrain regions activated a VGLUT1-positive glutamatergic pathway from the dorsal peduncular cortex and dorsal tenia tecta (DP/DTT), an unexplored prefrontal cortical area, to the dorsomedial hypothalamus (DMH), a hypothalamic autonomic center. Genetic ablation and optogenetics revealed that the DP/DTT→DMH pathway drives thermogenic, hyperthermic, and cardiovascular sympathetic responses to psychosocial stress without contributing to basal homeostasis. This pathway also mediates avoidance behavior from psychosocial stressors. Given the variety of stress responses driven by the DP/DTT→DMH pathway, the DP/DTT can be a potential target for treating psychosomatic disorders.

Psychological stressors induce various physiological responses by stimulating sympathetic, neuroendocrine, and behavioral mechanisms. Stress-induced sympathetic stimulation of thermogenic and cardiovascular functions is common in mammals and is recognized as a stress-coping response to boost physical performance in fight-or-flight situations. However, excessive stress may cause aberrant sympathetic symptoms such as psychogenic fever, a hyperthermia caused by intense psychological stressors (1). Sympathetic thermogenesis in brown adipose tissue (BAT) contributes to stress-induced hyperthermia in rats and humans (2, 3). Resistance of psychogenic fever to antipyretics indicates an etiology different from that of infection-induced fever (1). Patients with posttraumatic stress disorder exhibit augmented cardiovascular responses to stress, which increase the risk of developing hypertension and cardiovascular diseases (4). Vigorous psychogenic cardiovascular responses are also observed in panic disorder, a severe anxiety disorder (5). However, the central circuit mechanisms by which psychological stress activates the sympathetic nervous system are not well understood.

The dorsomedial hypothalamus (DMH) is an important brain site for sympathetic stress responses (6, 7). A hypothalamo-medullary pathway from the DMH to sympathetic premotor neurons in the rostral medullary raphe region (rMR) mediates BAT thermogenic and cardiovascular responses to psychological stress (2, 8). However, the mechanism by which the corticolimbic circuits that process stress and emotion activate the sympathoexcitatory DMH neurons remains unknown. We sought to identify the forebrain neural pathways that transmit stress signals to the DMH by using functional neuroanatomy and in vivo physiological experiments combining genetic lesion and optogenetic manipulations of a pathway of interest. We also employed social defeat stress (SDS), an animal model of psychosocial stress caused by interindividual relationships that mimics human social stress (9).

SDS activates DMH-projecting neurons in the mPFC

To label neurons that project to the DMH, we injected cholera toxin b subunit (CTb), a retrograde neural tracer, into the rat DMH and consequently found many CTb-labeled neurons in the medial prefrontal cortex (mPFC) (Fig. 1, A to C). Subsequent exposure to SDS significantly increased expression of Fos, a marker for neuronal activation, in CTb-labeled neurons in specific mPFC subregions: the prelimbic cortex (PrL), infralimbic cortex (IL), and dorsal peduncular cortex (DP) and the dorsal tenia tecta (DTT) immediately rostral to the callosal commissure (Fig. 1, D to F). These double-labeled cells were mostly distributed in layers V and VI and showed the typical morphology of glutamatergic pyramidal neurons (Fig. 1E). We did not find other forebrain regions containing a substantial double-labeled population after SDS.

Fig. 1 SDS activates DMH-projecting neurons in the mPFC.

(A) CTb injection into the DMH. 3V, third ventricle; f, fornix; mt, mammillothalamic tract; opt, optic tract. Scale bar, 500 μm. (B) CTb injection sites in the rats for which data are plotted in (F). VMH, ventromedial hypothalamic nucleus. (C) CTb-labeled neurons (arrowheads) in the mPFC. fmi, forceps minor of the corpus callosum; LV, lateral ventricle. Scale bars, 500 μm (main panel); 30 μm (inset). (D) Distribution of CTb-labeled cells with or without Fos immunoreactivity in the mPFC after sham handling (Control) or SDS (Stress). ac, anterior commissure; Acb, nucleus accumbens; Cg, cingulate cortex; E, ependymal and subependymal layer; M2, secondary motor cortex; Nv, navicular nucleus; OV, olfactory ventricle. (E) Stress-induced Fos expression (blue-black) in CTb-labeled (brown) cells in the DP (arrows). Note the typical pyramidal neuron morphology. Scale bars, 200 μm (main panel); 30 μm (inset). (F) Percentages of Fos-immunoreactive cells in CTb-labeled populations in mPFC subregions (n = 4 rats per group). *P < 0.05; **P < 0.01; ***P < 0.001 (unpaired t test; PrL, t6 = 6.16; IL, t6 = 5.93; DP, t6 = 7.41; DTT, t6 = 2.47). Error bars indicate SEM.

The DP/DTT mediates thermogenic and cardiovascular sympathetic responses to stress

The PrL and IL, which constitute the main body of the mPFC (10), exert inhibitory effects on stress responses (11, 12). Correspondingly, optogenetic stimulation of the IL→amygdala or PrL→bed nucleus pathway reduces freezing behavior in response to stress (13, 14). No literature is available on the function of the DP and DTT (hereafter, the DP/DTT). Therefore, we determined the roles of the stress-activated mPFC subregions in driving sympathetic stress responses.

SDS induced immediate increases in interscapular BAT temperature (TBAT) and body core (abdominal) temperature (Tcore) (Fig. 2), representing psychosocial stress–induced BAT thermogenesis and hyperthermia, respectively (8). These responses were abolished by inactivation of DP/DTT neurons with bilateral nanoinjections of muscimol: Responses were nearly suppressed by injections of 200 nl per site (TBAT: 80% inhibition, Tcore: 88% inhibition; Fig. 2, A to F) and were moderately inhibited by injections of 100 nl per site (TBAT: 63% inhibition, Tcore: 54% inhibition; fig. S1). In contrast, muscimol injections of 200 nl per site into the IL resulted in significantly reduced inhibition of the responses (TBAT: 33% inhibition, Tcore: 26% inhibition; Fig. 2, G to L) compared with injections into the DP/DTT (unpaired t tests comparing inhibition percentage; TBAT: t8 = 3.16, P < 0.05; Tcore: t8 = 2.82, P < 0.05). SDS also induced increases in heart rate (HR) and mean arterial pressure (MAP), which were diminished by injections of muscimol (100 nl per site) into the DP/DTT (HR: 53% inhibition, MAP: 77% inhibition; Fig. 2, M to Q).

Fig. 2 Neurons in the DP/DTT, rather than the IL, mediate sympathetic stress responses.

(A to Q) Effects of bilateral muscimol nanoinjections into the DP/DTT [(A) to (F) and (M) to (Q)] or the IL [(G) to (L)] on SDS-evoked BAT thermogenesis and hyperthermia [(A) to (L)] or cardiovascular responses [(M) to (Q)]. After saline or muscimol injections [1 mM; (A) to (L): 200 nl per site, (M) to (Q): 100 nl per site; marked by green microspheres in (A) and (G); scale bars, 500 μm] into the DP/DTT [(A), (B), and (M)] or the IL [(G) and (H)] (the right side of the symmetric bilateral injections is shown), animals were exposed to SDS [gray zones in (C), (E), (I), (K), (N), and (P)]. All rats (n = 5 per group) underwent two stress trials, >1 week apart, with saline (first trial) and muscimol (second trial) injections at the same sites. Time-course changes in TBAT, Tcore, HR, and MAP were analyzed by repeated measures two-way analysis of variance (ANOVA) [(C): injectant: F1,4 = 19.85, P = 0.011, time: F180,720 = 8.976, P < 0.001, interaction: F180,720 = 8.68, P < 0.001; (E): injectant: F1,4 = 36.89, P = 0.004, time: F180,720 = 10.93, P < 0.001, interaction: F180,720 = 10.38, P < 0.001; (I): injectant: F1,4 = 22.5, P = 0.009, time: F180,720 = 25.47, P < 0.001, interaction: F180,720 = 4.63, P < 0.001; (K): injectant: F1,4 = 0.35, P = 0.585, time: F180,720 = 12.64, P < 0.001, interaction: F180,720 = 2.07, P < 0.001; (N): injectant: F1,4 = 27.99, P = 0.006, time: F180,720 = 15.54, P < 0.001, interaction: F180,720 = 2.48, P < 0.001; (P): injectant: F1,4 = 55.59, P = 0.002, time: F180,720 = 11.16, P < 0.001, interaction: F180,720 = 2.47, P < 0.001] followed by Bonferroni’s post hoc test (red bars with asterisks indicate time points with significant differences). Area under the curve (AUC) during the stress period was analyzed by paired t test [(D): t4 = 7.57; (F): t4 = 5.93; (J): t4 = 3.46; (L): t4 = 1.39; (O): t4 = 4.39; (Q): t4 = 6.92]. *P < 0.05; **P < 0.01; ***P < 0.001. Error bars indicate SEM. bpm, beats per minute.

DP/DTT neurons drive sympathetic responses through activation of the DMH→rMR pathway

To test whether DP/DTT neurons transmit stress signals to the DMH to drive sympathetic responses, we performed in vivo electrophysiological experiments in anesthetized rats. Stimulation of DP/DTT neurons with nanoinjection of bicuculline robustly increased BAT sympathetic nerve activity (SNA), TBAT, HR, and MAP (fig. S2, A to E) (15). All of these responses were reversed by subsequent inactivation of DMH neurons with muscimol injections (fig. S2, A to E).

We next investigated whether the stimulation of DP/DTT neurons activated the DMH→rMR pathway. CTb injection into the rMR (fig. S2F) retrogradely labeled many neurons that bilaterally clustered in the dorsal part of the DMH (fig. S2, H and I); these include sympathoexcitatory glutamatergic neurons driving cold-defensive, febrile, and stress responses through their innervation of sympathetic premotor neurons in the rMR (8, 1618). Subsequent stimulation of DP/DTT neurons with a unilateral bicuculline injection under free-moving conditions increased Fos expression in the CTb-labeled population in the DMH, mainly ipsilaterally to the site of bicuculline injection (fig. S2, G to J).

Glutamatergic innervation of DMH neurons by DP/DTT neurons

Next, we determined the neurotransmitter that mediates the DP/DTT→DMH transmission. Transduction of DP/DTT neurons with palGFP, a membrane-targeted form of green fluorescent protein (GFP) (19), using adeno-associated virus (AAV) (Fig. 3A) allowed us to visualize their numerous axons projecting to the DMH. We found close association of these axons with DMH neurons retrogradely labeled with CTb from the rMR (Fig. 3, B and C, and fig. S3A), indicating putative synapses formed by DP/DTT-derived axons on DMH→rMR projection neurons. Confocal imaging revealed that the swellings of DP/DTT-derived axons apposed to rMR-projecting DMH neurons contained VGLUT1 (vesicular glutamate transporter 1), a marker for cortical glutamatergic neurons, but not VGLUT2, a subcortical glutamatergic neuron marker, or VGAT (vesicular GABA transporter), a GABAergic neuron marker (Fig. 3, D to F, and fig. S3B).

Fig. 3 Optogenetic stimulation of DP/DTT→DMH glutamatergic transmission elicits sympathetic responses.

(A and B) Injections of AAV-CMV-palGFP into the DP/DTT (A) and CTb into the rMR (B). The inset in (A) shows a palGFP-expressing neuron. py, pyramidal tract; RMg, raphe magnus nucleus; rRPa, rostral raphe pallidus nucleus. Scale bars, 500 μm (main panels); 20 μm (inset). (C) palGFP-labeled axons closely associated with a DMH neuron retrogradely labeled with CTb. Scale bar, 10 μm. (D to F) Pseudocolored confocal images showing palGFP-labeled axon swellings (arrowheads) apposed to CTb-labeled cell bodies (asterisks) in the DMH. VGLUT1 (D) (filled arrowheads), but not VGLUT2 (E) or VGAT (F) (hollow arrowheads), was detected in these axon swellings. See also fig. S3. Scale bars, 5 μm. (G) In vivo optogenetic stimulation of DP/DTT→DMH nerve endings. (H and I) DP/DTT neurons transduced with ChIEF-tdTomato (H) and their axons containing ChIEF-tdTomato in the DMH [(I); immunoperoxidase staining]. Scale bars, 500 μm. (J) Representative example of BAT thermogenic and cardiovascular responses elicited by bilateral laser illumination of DP/DTT→DMH nerve endings with ChIEF-tdTomato. AP, arterial pressure. (K) Changes in physiological variables induced by illumination of selective pathways (one-way ANOVA followed by Bonferroni’s post hoc test, n = 5 per group; ΔBAT SNA: F3,16 = 28.74, P < 0.001; ΔTBAT: F3,16 = 30.46, P < 0.001; ΔHR: F3,16 = 17.27, P < 0.001; ΔMAP: F3,16 = 8.29, P = 0.002). *P < 0.05; **P < 0.01; ***P < 0.001. See also fig. S4. (L) Experiment to test the effects of blockade of glutamatergic synapses in the DMH on physiological responses to photostimulation of DP/DTT→DMH transmission. (M) Each circle indicates a site of saline and AP5/CNQX injections made at the same location in the DMH of each rat (saline was always injected first). The right side of symmetric bilateral injections is shown. (N) AP5/CNQX injections into the DMH eliminated sympathetic responses elicited by photostimulation of DP/DTT→DMH nerve endings (compare with fig. S4F, which shows a result for the saline control). (O) Changes in physiological variables elicited by photostimulation of the DP/DTT→DMH pathway after saline or AP5/CNQX injections into the DMH (n = 5 per group). *P < 0.05; ***P < 0.001 (paired t test; ΔBAT SNA: t4 = 13.99; ΔTBAT: t4 = 3.98; ΔHR: t4 = 4.07; ΔMAP: t4 = 2.86). Error bars indicate SEM.

Consistent with the view that glutamatergic inputs to the DMH mediate stress responses, blockade of glutamatergic synapses in the DMH with nanoinjections of a mixture of 2-amino-5-phosphonovaleric acid and cyanquixaline (AP5/CNQX) inhibited SDS-induced sympathetic responses for the first 20 min of stress exposure (TBAT: 88% inhibition, Tcore: 80% inhibition, HR: 56% inhibition, MAP: 41% inhibition; fig. S3, C to F), after which the antagonistic effects of the drugs waned. The difference in the inhibitory effects of AP5/CNQX on the thermal and cardiovascular responses may be because the distribution of cardiovascular neurons in the DMH is broader than that of thermoregulatory neurons (20).

In vivo optogenetic stimulation of DP/DTT→DMH glutamatergic transmission elicits sympathetic responses

To directly examine the sympathoexcitatory role of the DP/DTT→DMH pathway, we selectively stimulated DP/DTT→DMH monosynaptic transmission by using an in vivo optogenetic technique (Fig. 3G). ChIEF, a channelrhodopsin variant (21), was used to induce membrane depolarization and action potentials in neurons with exposure to light. AAV transduction of DP/DTT neurons with ChIEF-tdTomato or palGFP resulted in the localization of the expressed proteins in their cell bodies in the DP/DTT and nerve endings densely distributed in the DMH (Fig. 3, H and I, and fig. S4, A and B). Illumination of ChIEF-tdTomato–containing nerve endings, but not palGFP-containing nerves, in the DMH with pulsed blue laser light for 30 s consistently increased BAT SNA, TBAT, HR, and MAP in anesthetized rats (Fig. 3, J and K, and fig. S4C). Longer illumination (180 s) elicited larger increases, and the photostimulated BAT thermogenesis was as intense as that induced by SDS (fig. S4, D and E, and table S1). Photostimulation of DP/DTT→DMH transmission activated the DMH→rMR pathway (fig. S4, G to I). Consistent with the view that DP/DTT→DMH transmission driving the responses is glutamatergic, antagonizing glutamate receptors in the DMH eliminated all sympathetic responses elicited by photostimulation of DP/DTT→DMH transmission (Fig. 3, L to O, and fig. S4F).

DP/DTT-derived nerves with ChIEF-tdTomato were also distributed in the lateral hypothalamic area (LH) (Fig. 3I), which contains orexin neurons proposed to participate in sympathetic stress responses (22). However, photostimulation of DP/DTT→LH transmission elicited no or subtle changes in the physiological variables (Fig. 3K and fig. S4J). Additionally, no response was elicited by photostimulation of IL→DMH axonal projections with ChIEF-tdTomato (Fig. 3K and fig. S4, K to M).

DP/DTT→DMH transmission is essential for sympathetic responses to psychosocial stress

We next employed genetic approaches to determine the importance of the DP/DTT→DMH monosynaptic pathway for driving sympathetic stress responses. First, DP/DTT→DMH projection neurons were selectively lesioned. An AAV for Cre recombinase–dependent expression of taCasp3, which induces cell-autonomous apoptosis (23), and a retrograde AAV (AAVrg) for Cre expression were bilaterally injected into the DP/DTT and DMH, respectively, to selectively lesion double-infected neurons (i.e., DP/DTT→DMH projection neurons) (Fig. 4A). Figure 4B shows an example of lesions of DP/DTT→DMH neurons: taCasp3 expression eliminated 88% of Cre-expressing (i.e., DMH-projecting) DP/DTT neurons in one representative rat as compared with palGFP-transduced control rats. This lesioned rat failed to exhibit BAT thermogenic and hyperthermic responses to SDS (Fig. 4C). Group data (Fig. 4D and fig. S5A) showed that stress-induced BAT thermogenesis was inhibited in proportion to the extent of lesions of DP/DTT→DMH neurons, and the stress response was totally eliminated when all the neurons were lesioned. However, the lesions of DP/DTT→DMH neurons did not affect diurnal control of TBAT or Tcore (fig. S5B).

Fig. 4 Selective lesion or inhibition of the DP/DTT→DMH pathway abolishes sympathetic stress responses.

(A) Scheme for genetic lesions of DP/DTT→DMH neurons. ITR, inverted terminal repeat; WPRE, woodchuck hepatitis virus posttranscriptional regulatory element; bGH polyA, bovine growth hormone polyadenylation signal. (B and C) Representative examples of palGFP or taCasp3 transduction of DP/DTT→DMH neurons from a single rat. (B) Cre-immunopositive (DMH-projecting) neurons were eliminated exclusively in the DP/DTT after taCasp3 transduction. Arrowheads indicate deep layers of the DP/DTT. Scale bar, 500 μm. (C) This lesioned rat failed to exhibit thermal responses to SDS (gray zone). (D) Relationship between the number of Cre-positive neurons remaining in the DP/DTT and stress responses (AUC of ΔTBAT and ΔTcore during stress; see also time-course data in fig. S5A). Data from taCasp3 rats were subjected to linear regression analysis (Pearson’s correlation test). The averages (AVG) of AUC were compared for the palGFP and taCasp3 groups (unpaired t test; n = 5 per group; ΔTBAT: t8 = 4.30; ΔTcore: t8 = 5.81). **P < 0.01; ***P < 0.001. Arrows indicate the cases plotted in (C). (E) In vivo optogenetic inhibition of DP/DTT→DMH neurons. Many Cre-immunopositive (DMH-projecting) DP/DTT neurons expressed iChloC-mCherry. Scale bar, 20 μm. (F) DP/DTT neurons were transduced with palGFP or iChloC-mCherry (top), and their axons with the proteins were densely distributed in the DMH (bottom; immunoperoxidase staining). Scale bars, 500 μm (top); 20 μm (bottom). (G) Illumination of DP/DTT→DMH nerve endings with iChloC-mCherry for the first 10 min (blue zone) of SDS inhibited cardiovascular and hyperthermic stress responses. Increases in HR and MAP after SDS were caused by stress from returning to the home cage. Time-course changes in the variables were analyzed by repeated measures two-way ANOVA (n = 5 per group; ΔHR: group: F1,8 = 23.7, P = 0.001, time: F180,1440 = 23.55, P < 0.001, interaction: F180,1440 = 5.68, P < 0.001; ΔMAP: group: F1,8 = 5.98, P = 0.040, time: F180,1440 = 11.82, P < 0.001, interaction: F180,1440 = 2.24, P < 0.001; ΔTcore: group: F1,8 = 9.69, P = 0.014, time: F180,1440 = 31.83, P < 0.001, interaction: F180,1440 = 7.69, P < 0.001) followed by Bonferroni’s post hoc test (red bars with asterisks indicate time points with significant differences). The difference in AUC during the stress period was analyzed by unpaired t test (ΔHR: t8 = 8.21; ΔMAP: t8 = 4.91; ΔTcore: t8 = 3.91). *P < 0.05; **P < 0.01; ***P < 0.001. Error bars indicate SEM.

To exclude the possible influence of long-term alterations in the circuits after neuronal lesions, we optogenetically suppressed DP/DTT→DMH transmission in free-moving rats. The Cre-dependent expression system with anterograde and retrograde AAVs was used to selectively transduce DP/DTT→DMH projection neurons with iChloC, a chloride-conducting channelrhodopsin shown to photoinhibit neuronal activity (24) (Fig. 4E). Expressed iChloC-mCherry proteins (or palGFP for control) were bilaterally detected in cell bodies in the DP/DTT and their nerve endings in the DMH (Fig. 4F). In vivo bilateral illumination of iChloC-mCherry–containing nerve endings in the DMH with pulsed blue light exerted potent, long-lasting inhibition on SDS-induced increases in HR, MAP, and Tcore (HR: 63% inhibition, MAP: 52% inhibition, Tcore: 66% inhibition; Fig. 4G) as compared with illumination of palGFP-expressing nerves. Commensurate with these inhibitions, iChloC-mCherry was expressed in 70 ± 2% of Cre-expressing (DMH-projecting) DP/DTT neurons (counted in three rats in which Cre immunostaining was successful; Fig. 4E). Photoinhibition of DP/DTT→DMH transmission did not reduce HR, MAP, or Tcore below baseline levels, which indicates that this pathway is not involved in basal maintenance of these parameters.

Photoinhibition of DP/DTT→DMH transmission also suppressed BAT thermogenesis evoked by cage-exchange stress, an anticipatory anxiety model, in males and females (fig. S6), indicating that DP/DTT→DMH transmission mediates sympathetic responses to broader types of psychological stressors in both sexes. In contrast, photoinhibition of IL→DMH transmission did not impede SDS-induced stress responses (fig. S7).

The DP/DTT→DMH pathway mediates stress behaviors and skin vasoconstriction

Because the DP/DTT→DMH pathway is essential for driving the repertoire of sympathetic stress responses, we sought to examine its involvement in stress behavior. We thus investigated the effect of optogenetic inhibition of DP/DTT→DMH transmission on avoidance behavior from psychosocial stressors (25). A male Wistar rat in which DP/DTT→DMH neurons were selectively transduced with iChloC-mCherry or palGFP (Fig. 4E) was subjected to SDS and moved to an open field for habituation (Fig. 5A). Subsequently, the dominant male Long-Evans rat used in the SDS episode was caged and placed in the open field (Fig. 5A). All of the stressed Wistar rats that did not undergo illumination exhibited avoidance behavior by staying away from the social interaction zone surrounding the cage of the Long-Evans rat (Fig. 5, B and C). By contrast, naïve Wistar rats, which had not experienced SDS, exhibited active social interactions (Fig. 5C). The stress-induced avoidance behavior was suppressed by photoinhibition of DP/DTT→DMH transmission in iChloC-mCherry rats, and their social interactions were restored to the level exhibited by naïve rats, whereas illumination in palGFP rats had no effect (Fig. 5, B and C). Illumination administered when iChloC-mCherry rats were lying down in a relaxed state in the open field without the presence of a Long-Evans rat did not elicit locomotion (trials in four rats, five illumination episodes per rat). Therefore, it is unlikely that the behavioral effect of photoinhibition resulted from stimulation of any motivation or vigilance circuit system. In addition, photoinhibition did not alter social interactions exhibited by naïve rats (naïve ON; Fig. 5C).

Fig. 5 Photoinhibition of the DP/DTT→DMH pathway abolishes stress-induced avoidance behavior and skin vasoconstriction.

(A) Wistar rats in which DP/DTT→DMH neurons were transduced with iChloC-mCherry or palGFP (Fig. 4E) underwent a social interaction test after SDS. (B) Representative examples of behavior tracking of palGFP- and iChloC-mCherry–transduced rats during the social interaction test. (C) Behaviors during the social interaction test (n = 5 per group). Naïve rats did not undergo SDS before habituation and social interaction test. Data were analyzed by one-way ANOVA (Entries: F5,24 = 12.87, P < 0.001; Time spent: F5,24 = 15.49, P < 0.001; Moving distance: F5,24 = 9.15, P < 0.001) followed by Bonferroni’s post hoc test. ns, not significant; **P < 0.01; ***P < 0.001. Error bars indicate SEM. (D and E) Thermographic measurements at proximal and distal parts [arrowheads in (D)] on the tail skin (n = 5 per group) were analyzed by repeated measures two-way ANOVA (Proximal: treatment: F1,4 = 17.44, P = 0.014; group: F1,4 = 2.11, P = 0.220; interaction: F1,4 = 9.01, P = 0.040; Distal: treatment: F1,4 = 12.13, P = 0.025; group: F1,4 = 10.74, P = 0.031; interaction: F1,4 = 11.36, P = 0.028) followed by Bonferroni’s post hoc test. ns, not significant; *P < 0.05 (versus habituation). Error bars indicate SEM. (F) Model of the central psychosomatic pathways that drive stress responses. The DP/DTT integrates signals from multiple forebrain regions processing stress and emotion and then provides a glutamatergic (Glu) master signal to the DMH to excite neuronal groups controlling different effectors. The excited DMH neurons drive sympathetic outflows to BAT, the heart, and cutaneous and visceral blood vessels through the rMR and partly through the rostral ventrolateral medulla (RVLM). Another subset of DMH neurons drives avoidance behavior through as-yet-unknown circuit mechanisms. AIP, posterior part of the agranular insular cortex; Pir2, layer II of the piriform cortex.

Thermography revealed reduction of tail skin temperature in palGFP rats during the social interaction test (with illumination) compared with temperature during habituation (Fig. 5, D and E). This outcome represents stress-induced cutaneous vasoconstriction, a sympathetic response that contributes to stress-induced hyperthermia by reducing heat loss (26). Tail skin vasoconstriction was absent in iChloC-mCherry rats with photoinhibition of DP/DTT→DMH transmission (Fig. 5, D and E).

The DP/DTT receives stress-driven inputs from thalamic and cortical regions

To explore the forebrain regions that provide stress inputs to the DP/DTT, we combined retrograde tracing from the DP/DTT with detection of stress-induced Fos expression. Rats that received a CTb injection into the DP/DTT (fig. S8, A and B) underwent SDS or sham handling. Substantial numbers of neurons were labeled with CTb in several regions of the thalamus, insular and piriform cortices, and amygdala. SDS significantly increased Fos expression in CTb-labeled neurons in the mediodorsal (MD) and paraventricular (PVT) thalamic nuclei, the posterior part of the agranular insular cortex, and layer II of the piriform cortex (fig. S8, C to F). This observation suggests that the DP/DTT receives and integrates stress-driven inputs from these thalamic and cortical regions and provides the integrated signals to the DMH to drive sympathetic and behavioral stress responses (Fig. 5F).


By using the rodent model of psychosocial stress, we discovered a prefrontal cortex–hypothalamus excitatory pathway that drives sympathetic and behavioral stress responses, in which the DP/DTT was crucial for stress signaling to the DMH. Our anatomical, physiological, and optogenetic experiments revealed that VGLUT1-positive pyramidal neurons in the DP/DTT transmit psychological stress–driven glutamatergic signals to the DMH to elicit a variety of stress responses. Most notably, selective ablation or inhibition of the DP/DTT→DMH monosynaptic pathway abolished BAT thermogenic, skin vasoconstrictor, cardiovascular, and behavioral responses to SDS without affecting basal thermoregulatory or cardiovascular homeostasis. Our present findings demonstrate that the DP/DTT→DMH excitatory transmission of psychological stress signals is a master driver of the wide range of sympathetic and behavioral stress responses (Fig. 5F).

The DP/DTT is an unexplored brain area located at the ventral limit of the mPFC. In contrast, the PrL and IL have been a central focus of stress research and have been shown to provide signals to inhibit stress responses. In this study, we discovered a group of neurons that drive stress responses in the DP/DTT. Photostimulation of DP/DTT→DMH transmission, but not that of IL→DMH, elicited sympathetic responses that mimic stress responses. Photoinhibition of DP/DTT→DMH transmission, but not that of IL→DMH, suppressed sympathetic stress responses. Also, inactivation of DP/DTT neurons suppressed stress responses, in contrast to subtle effects of inactivation of IL neurons. Thus, we propose that there are two functional units in the mPFC: the ventral (DP/DTT) unit that drives stress responses and the dorsal (PrL/IL) unit that inhibits these responses. The inhibitory unit may constitute the negative feedback mechanism in which the stress hormones, glucocorticoids, act in the PrL and IL to mitigate or terminate stress responses (11). This feedback inhibition might involve stress-activated PrL/IL→DMH neurons (Fig. 1) and/or local inhibition of DP/DTT neurons from the PrL/IL.

DP/DTT→DMH stress signaling activates DMH→rMR sympathoexcitatory neurons to drive stress responses (Fig. 5F). The DMH→rMR pathway also serves as the trunk pathway that controls body temperature and develops inflammatory fever, whose activity level is continuously controlled by descending inputs from the thermoregulatory center, the preoptic area (27). Because the DP/DTT→DMH pathway does not contribute to basal thermoregulation or cardiovascular control, the stress-driven ad hoc inputs from the DP/DTT are likely integrated in the DMH with the homeostatic tonic inputs from the preoptic area by impinging on the DMH→rMR efferent neurons. In addition, stress-induced visceral and cutaneous vasoconstriction may also be mediated in part by DMH neurons innervating the rostral ventrolateral medulla (28, 29) (Fig. 5F).

The DP/DTT→DMH pathway appears to constitute a key psychosomatic connection through which stress and emotions affect the autonomic and behavioral motor systems. Although the corticolimbic circuits that process stress and emotions are undetermined, the PVT and MD thalamic nuclei, which provide stress inputs to the DP/DTT, constitute a fear stress circuit involving the amygdala (30, 31). Thus, the stress and emotion signals processed by forebrain circuits are likely integrated at the DP/DTT and then transmitted to the DMH. In panic disorder, glutamatergic inputs to the DMH to develop the panic-prone state (32) may be provided from the DP/DTT. The DP/DTT→DMH pathway is a potential target for treating psychosomatic disorders that involve aberrant physiological responses, particularly because this pathway does not contribute to basal autonomic homeostasis.

Supplementary Materials

Materials and Methods

Figs. S1 to S8

Table S1

References (3338)

References and Notes

Acknowledgments: We thank P. Aebischer, P. Hegemann, A. Y. Karpova, R. McQuiston, D. V. Schaffer, N. Shah, J. Wells, and J. Wietek for sharing their plasmids and H. Hioki for providing anti-VGLUTs antibodies. Funding: This study was supported by the JSPS NEXT program (LS070 to Ka.N.); MEXT KAKENHI (16H05128, 15H05932, 26118508, 26713009, and 22689007 to Ka.N.; 19K06954 and 16K19006 to N.K.); JST PRESTO (JPMJPR13M9 to Ka.N.); AMED (JP19gm5010002 to Ka.N.); and grants from the Takeda Science Foundation, Nakajima Foundation, Uehara Memorial Foundation, Ono Medical Research Foundation, Brain Science Foundation, Kowa Life Science Foundation (to Ka.N.), and Kato Memorial Bioscience Foundation (to N.K.). Author contributions: Ka.N. conceived the project. N.K. and Ka.N. designed experiments. All authors performed experiments and analyzed data. N.K. and Ka.N. discussed data and wrote the manuscript. Ka.N. supervised the entire project. All authors approved the manuscript. Competing interests: The authors declare no competing interests. Data and materials availability: All data are available in the main text or the supplementary materials. Plasmids and AAVs with Addgene numbers given in the materials and methods were obtained under material transfer agreements with Addgene.

Stay Connected to Science

Navigate This Article